C2C Distance Interpretations

[civengrgmu]

Good afternoon,

I am currently looking at some C2C distances between two point clouds of a highway bridge. The difference in the point clouds is that the second is loaded with a known weight - and I'm trying to determine the amount that the loaded bridge is deflecting by looking at changes between the point clouds.

I have the coordinate axis set so that the Z-axis is aligned with the direction of the loads (i.e., the direction of the deflection that I'm trying to measure), such that the deformations occur in the "negative" Z-direction. I set the "undeformed" point cloud as the "reference" cloud, and I set the "deformed" point cloud as the "compared" cloud.

Does the C2C distance measure the distance "from the reference to the compared" cloud, or is it the other way around?

The reason that I ask is because I am getting a distribution of values (as expected) where some values are positive and some are negative. Based on the deflection magnitudes that we are seeing and the quality of the point clouds that we are working with for this experiment, this is fine - but I want to make sure that I correctly understand the output that I'm seeing from CloudCompare.

Of note - this question is specifically about the C2C distance comparison, and not the M3C2 comparison (the M3C2 comparison is another discussion for me...).

Thanks.

[daniel]

So I guess you are splitting the resulting distances as 3 scalar fields along X, Y and Z? (as that's the only way to get signed values with this tool).

Anyway, the values should be 'compared point.z - reference point.z' if I'm not mistaken (see https://github.com/CloudCompare/CCCoreL ... s.cpp#L486, where the query point is a point of the compared cloud, and 'P' is the nearest point found in the reference cloud).

And I could have suggested to use M3C2, but actually another option is to use the 'Tools > Volume > Compute 2.5D Volume' tool (where you can get a direct comparison map along Z, very easily, and you can export the resulting grid as a cloud).

[civengrgmu]

Thanks Daniel. And yes - I'm splitting the resulting distances into the 3 component scalar fields. I'll give the 2.5D volume tool a try, as well.